Design And Implementation Of Eye Tracking System Based On FPGA

Eye tracking technology is an important branch of machine vision. The mainresearch is feature detection and recognition of human eye，then use the detection resultto calculate the eye fixation. In neuroscience, psychology, human-computer interaction,ergonomics, clinical diagnosis and other fields, eye-tracking technology has a broadapplication prospects. The design method and principle of eye-tracking systemdetermines its performance. Among these eye tracking systems, the one use pupilcorneal reflection principle has high accuracy of eye fixation locating, and the user doesnot need to wear a helmet. So the principle has been widely applied.In this dissertation, a non-contact eye tracking system is designed and implementedbased on the pupil corneal reflection principle, digital image processing, SOPC andother technology. The system is mainly used in human computer interaction in theoffice environment. According to the function requirement, this paper puts forward thedesign scheme of the system, and carry out the system design from hardware andsoftware.The system uses DE2development board as hardware platform. And the hardwaredesign mainly includes the infrared LED light source module, image acquisition module,SOPC system, VGA display module, MP3module. LED light source provides goodillumination environment to enhance the anti disturb ability. Image acquisition moduledecodes the analog video signal captured by a camera and finally obtains the gray levelimage. The SOPC system takes Nios II processor as core and integrates manyperipherals. And it is the main task of hardware design. VGA display module usesSRAM to store the image data. In order to overcome the contradiction between thememory capacity and display effect, format of memory data is defined. A custom VGAcontroller is used to decode the image data and generate signals to drive the monitor.MP3communicates with FPGA through UART serial port, and it is the main peripheralfor voice helping function.The main task of the software design is image processing. At first, an eye locationalgorithm is used to search the eye, and a local eye image is obtained. The eye locationalgorithm makes full use of the image gradient, gray and eye position information, andit is fast and accurate. Then threshold segmentation, edge detection, circle fitting andother algorithms were used to locate pupil center and corneal reflection spot center form the local eye image. After that, the mapping function between the pupil-spot relativeoffset and screen coordinates is acquired through calibration. According to the mappingfunction and pupil-spot relative offset, the eye fixation is calculated and superimposedon the VGA display. In addition, the eye controlled human-computer interaction isimplemented.Finally, the eye tracking system was tested. System test results show that, thissystem has realized the eye tracking function. It can also be applied to human-computerinteraction system and implements eye controlled human-computer interaction. Finally,the design results meet the functional requirements of the system.The eye tracking system has strong ability of resisting the ambient lightinterference. It can be used in the daytime and night. The system has the advantages ofsimple structure, low cost and high positioning accuracy. The hardware and softwarecan also be upgraded. In the aspect of eye controlled human-computer interactionresearch, this system has some reference value.